Low Temperature Mitigates Cardia Bifida in Zebrafish Embryos

The coordinated migration of bilateral cardiomyocytes and the formation of the cardiac cone are essential for heart tube formation. We investigated gene regulatory mechanisms involved in myocardial migration, and regulation of the timing of cardiac cone formation in zebrafish embryos. Through screening of zebrafish treated with ethylnitrosourea, we isolated a mutant with a hypomorphic allele of mil (s1pr2)/edg5, called s1pr2as10 (as10). Mutant embryos with this allele expressed less mil/edg5 mRNA and exhibited cardia bifida prior to 28 hours post-fertilization. Although the bilateral hearts of the mutants gradually fused together, the resulting formation of two atria and one tightly-packed ventricle failed to support normal blood circulation. Interestingly, cardia bifida of s1pr2as10 embryos could be rescued and normal circulation could be restored by incubating the embryos at low temperature (22.5°C). Rescue was also observed in gata5 and bon cardia bifida morphants raised at 22.5°C. The use of DNA microarrays, digital gene expression analyses, loss-of-function, as well as mRNA and protein rescue experiments, revealed that low temperature mitigates cardia bifida by regulating the expression of genes encoding components of the extracellular matrix (fibronectin 1, tenascin-c, tenascin-w). Furthermore, the addition of N-acetyl cysteine (NAC), a reactive oxygen species (ROS) scavenger, significantly decreased the effect of low temperature on mitigating cardia bifida in s1pr2as10 embryos. Our study reveals that temperature coordinates the development of the heart tube and somitogenesis, and that extracellular matrix genes (fibronectin 1, tenascin-c and tenascin-w) are involved.


Introduction
During zebrafish cardiac development, heart progenitors are located near the ventral lateral margins at 5 hours post-fertilization (hpf). By 12 hpf, both myocardial and endocardial progenitors appear bilaterally under the future hindbrain region. Bilateral cardiomyocytes then migrate toward the midline and fuse to form a cone-shaped heart at the 21-somite stage (ss). Next, the heart cone begins to elongate, and forms a single heart tube at 24 hpf [1][2][3].
Previous studies have demonstrated that mutations in several genes cause defective migration of myocardial precursors, including genes involved in myocardium formation [4,5], endoderm formation [5][6][7], extracellular matrix composition [8], and sphingosine-1-phosphate signaling [9][10][11]. Studies on fibronectin mutants of mice and chick embryos incubated with fibronectin antibodies demonstrated that fibronectin is the major extracellular matrix component that directs myocardial migration to the midline [12,13]. In zebrafish, fibronectin has been shown to be required for the proper formation of adherens junctions between myocardial precursor cells; this in turn maintains epithelial integrity, which is essential for myocardial migration [8]. Increased levels of fibronectin 1 (fn1) mRNA were previously observed in hand2 mutants; a genetic reduction of fn1 gene dosage in hand2 mutants partially rescued the cardia bifida phenotype, thereby implicating fibronectin in the development of this disorder [14].
The mil/edg5 gene encodes a G protein-coupled receptor which binds to sphingosine-1-phosphate in the lateral plate mesoderm [10], while toh/spns2 encodes the sphingosine-1-phosphate transporter in the yolk syncytial layer, required for transport of sphingosine-1-phosphate to the mesoderm [9,11]. Mutations in either gene have been reported to cause cardia bifida. Injection of fibronectin into the midline of mil 14-ss embryos could partially rescue the cardia bifida, and caused partial fusion of bilateral cardiomyocytes [15]. Decreased levels of fibronectin were also detected at the midline in embryos injected with edg5-morpholino antisense oligomers (MO) [11]. The findings of these studies indicate that the cardia bifida exhibited by mil mutant embryos can be partially attributed to decreased cell-fibronectin interactions. However, both the regulatory mechanisms that control genes involved in myocardial migration, and the developmental processes required for heart tube formation, remain unclear.
In this study, we screened zebrafish treated with ethylnitrosourea (ENU), and isolated a hypomorphic allele of mil (s1pr2)/ edg5, designated as s1pr2 as10 (as10). Embryos with the s1pr2 as10 allele exhibited decreased mil/edg5 expression, and this was associated with cardia bifida at 24 hpf. Interestingly, the cardia bifida phenotype of s1pr2 as10 embryos could be rescued (restoring normal circulation) by incubating the embryos at a low temperature (22.5uC). A similar rescue was observed in gata5 and bon cardia bifida morphants raised at 22.5uC. DNA microarray, MO knockdown, and rescue experiments with mRNA or protein revealed that low temperature mitigates cardia bifida by regulating the expression of extracellular matrix genes, particularly in the anterior lateral plate mesoderm, midline, and pharyngeal arch regions.

Ethics Statement
All animal procedures were approved by the Animal Use and Care Committee of Academia Sinica (Protocol # RFi-ZOOHS2010065).

Zebrafish Maintenance and Staging, and Low Temperature and N-acetyl Cysteine (NAC) Treatment
Adult zebrafish strains, including AB, SJD, Tg(27.5bmp4:GFP) as1 , Tg(cmlc2:EGFP, cmlc2:H2AFZmCherry) cy13 , s1pr2 as10 (as10), and mil (s1pr2 m93/+ )/edg5, were raised under standard conditions, as previously described [16]. To perform the rescue experiments, embryos were incubated at 22.5uC from the one-cell zygote stage onwards. Two staging systems for zebrafish embryos were used in this study, the somite stage (ss), which is determined by counting the somite number, and the conventional temporal stage, which is determined by hours postfertilization (hpf). We also followed the previously described morphological criteria for staging [17]. One-cell zygotes of as10 were incubated at 22.5uC and treated with 50 or 150 mM N-acetyl cysteine (NAC) (Merck) from the tailbud stage onwards. NACtreated embryos were continuously incubated at 22.5uC and allowed to develop to the 26 ss stage; the position of migrating cardiomyocytes was examined at this stage.
Positional Cloning for s1pr2 as10 s1pr2 as10 heterozygous mutant fish in the AB background were outcrossed to the wild-type SJD background to generate mapping stocks. Simple sequence length polymorphism (SSLP) markers were used to perform the linkage analyses. For the low-resolution mapping analysis, s1pr2 as10 was mapped to chromosome 3 between markers z3725 (9 recombinants out of 71 meioses, 9/71) and Z39291 (2/78). To construct a high-resolution map around the s1pr2 as10 locus, 900 s1pr2 as10 homozygous mutant embryos and custom-designed SSLP markers were used. The minimum recombinants included markers CU57-40022 (1/900) and CR38-83127 (2/900). The sequences of the primer pair for CU57-40022 were F-TCATATCGATTGACCACCCTAA and R-ATTATGGAGCAGCCGAGCTA. The sequences of the primer pair for CR38-83127 were F-GTCAGTGT-CACTCCCCTGGT and R-TGTGACCCATTCTTT-CATTTCTT.

Ligation-mediated Polymerase Chain Reaction (PCR)
Ligation-mediated PCR was performed to identify an insertion in intron 1 of the mil/edg5 gene. Genomic DNA isolated from s1pr2 as10 mutants was digested with NlaIII, and ligated to an Nla linker prepared by annealing Nla oligo 1 (GTAATACGACT-CACTATAGGGCTCCGCTTAAGGGACCATG) and Nla oligo 2 (GTCCCTTAAGCGGAG). The first PCR was conducted using the Nla linker-ligated genomic DNA of s1pr2 as10 mutants as a template with primer 1 (GTAATACGACTCACTATAGGGC) and primer 2 (AGAGTGGTCATGTGGTGGTC). A nested PCR reaction was subsequently performed using a 1:50 dilution of the first PCR product as a template with primer 1 (AGGGCTCCGCTTAAGGGAC) and primer 2 (CTGATAAG-TAAATGAATGAACACA). The candidate PCR products were subsequently gel-eluted and cloned into the pGEM-T vector (Promega). Sequence analyses indicated that the insertion was derived from chromosome 11. To further confirm the presence of the insertion in intron 1 of the mil/edg5 gene, genomic DNA was isolated from wild-type, the wild-type siblings of s1pr2 as10 , and s1pr2 as10 mutant embryos, and used as templates for PCR with primers either flanking or within the insertion ( Figure S1B). The sequences of the three primers were as follows: F1-AAAA-CATCTGCACGCGCTTCTTC; R1-ATAAGAGTGGT-CATGTGGTGG TC; and I1-GCGTGGCTCATTATACTT-CAGGA.
Whole-mount in Situ Hybridization, Histological Analysis, and Photography Embryos treated with 0.003% phenylthiocarbamide were subjected to whole mount in-situ hybridization, using digoxigeninor fluorescein-labeled antisense RNA probes and either alkaline phosphatase-conjugated anti-digoxigenin or anti-fluorescein antibodies, as previously described. Various templates derived from the pGEM-T vector were linearized, and the following antisense RNA probes were generated, using the restriction sites and promoters in parentheses: amhc (Nco I/sp6), cmlc2 (Nco I/sp6), foxa1 (Sal I/T7), fn1 (Nco I/sp6), tnc (Apa I/sp6), and tnw (Sac II/sp6). Paraffin sectioning (5 mm) and haematoxylin and eosin staining were performed using standard procedures. All images were taken using a Zeiss AxioCam HRC camera mounted on a Zeiss Imager M1 microscope.

Immunohistochemistry
Immunohistochemistry was performed as previously described [23]. In brief, embryos were fixed in 2% paraformaldehyde at 4uC overnight. Fixed embryos were embedded in 4% low-melting point agarose and sectioned with a vibratome (Leica VT1000M) to sections of 200 mm. The sections were washed with 1% Triton X-100 in PBS (PBSTx) at room temperature, before being incubated in blocking solution (10% lamb serum in PBSTx). Sections were then incubated with an anti-fibronectin antibody (Sigma) at a dilution of 1:100 for 1.5 days at 4uC. Alexa Fluor 488 goat antirabbit IgG (Invitrogen) was used as a secondary antibody, at a dilution of 1:200. The sections were washed with blocking solution and PBSTx at room temperature. After staining, sections were examined with a Leica Confocal Microscope (TCS SP5 MP).

DNA Microarray, Gene Ontology (GO), and Pathway Analyses
Total RNA was extracted from 22-ss wild-type and s1pr2 as10 mutant embryos raised at either 28.5 or 22.5uC using an RNeasy Plant Mini Kit (Qiagen), and 10 mg of RNA was treated with RNase-free DNase I. The RNA quality was verified with a Bioanalyzer 2100 (Agilent). DNA microarray analysis was performed by NimbleGen using Zebrafish Gene Expression 126135K arrays (NimbleGen) containing 38,489 genes. The 1,844 genes that exhibited a 1.3-fold or greater change in expression between embryos raised at 28.5 and 22.5uC for either s1pr2 as10 homozygous mutants or wild-type were clustered into 5 groups using GeneSpring GX software (Agilent). The same strategy was applied to the 264 genes that exhibited a 1.5-fold or greater change in expression. The identified genes were subjected to gene ontology (GO) analysis using a Batch-Genes tool based on a zebrafish or human database (GOEAST, http:// omicslab.genetics.ac.cn/GOEAST/tools.php), for which the p values were smaller than 0.1 [24]. Pathway analysis was performed on these genes using the website of the Database for Annotation, Visualization and Integrated Discovery (DAVID, http://david. abcc.ncifcrf.gov/), for which p values were smaller than 0.1 [25,26].

Digital Gene Expression (DGE) Analysis
Total RNA was extracted from 22-ss s1pr2 as10 mutant embryos raised at 28.5 or 22.5uC using an RNeasy Plant Mini Kit (Qiagen), and 4 mg was treated with RNase-free DNase I. The RNA quality was verified with a Bioanalyzer 2100 (Agilent). DGE analysis was performed by BGI through next-generation sequencing (NGS). In brief, Oligo(dT) beads were used to enrich mRNA from total RNA, and the mRNA was then reverse transcribed into double-stranded cDNA, using reverse transcriptase and DNA polymerase. The cDNA was digested with NlaIII and ligated to Illumina adaptor 1. The cDNA was then cut at 17 bp downstream of the CATG site using MmeI, and its 39 end was ligated to Illumina adaptor 2. PCR was performed using primers GX1 and GX2, which target adaptors 1 and 2, respectively. The resulting 95-bp fragments were isolated by 6% TBE polyacrylamide gel electrophoresis (PAGE), and the DNA was purified and subjected to Illumina sequencing using a Genome Analyzer II (Illumina).

Phenotypic Characterization of the s1pr2 as10 Mutation
In order to screen for mutants with heart morphogenetic defects, we conducted a ethylnitrosourea mutagenesis screen of the Tg(27.5bmp4:GFP) as1 transgenic zebrafish line, which expresses GFP in the heart after 24 hpf. Through this screen, we isolated the s1pr2 as10 mutant line ( Figure 1B), which is a recessive embryonic lethal mutation with several phenotypes, including pericardial edema (Figure 1 B'), small eyes, and tail blisters (Figure 1 B'). Blood circulation was not established during the embryonic and larval periods. We subsequently crossed s1pr2 as10 into a Tg(cmlc2:EGFP, cmlc2:H2AFZmCherry) cy13 background to generate higher fluorescence intensity in the bilateral cardiomyocytes. In contrast to wild-type (WT) myocardial precursors, which migrated toward the midline and fused to form a single heart tube by 24 hpf ( Figure 1C), myocardial precursors in the s1pr2 as10 mutants remained in the bilateral lateral plate mesoderm (LPM) at 24 hpf ( Figure 1D). Although the s1pr2 as10 bilateral myocardial precursors eventually reached the midline, they formed a threechambered heart with two atria and one tightly-packed ventricle from 48 hpf ( Figure 1F and 1H); this is in contrast to WT hearts, which contain a single atrium and ventricle ( Figure 1E and 1G). These results indicate that the newly identified s1pr2 as10 mutation delays the fusion of bilateral myocardiocytes, leading to the formation of non-functional three-chambered hearts.
Genetic Mapping of the s1pr2 as10 Mutant Gene By performing positional cloning with simple sequence length polymorphism markers, we established that s1pr2 as10 is located on chromosome 3 near the mil/edg5 locus ( Figure S1A). Sequencing of the mil/edg5 coding regions within cDNA or genomic DNA of s1pr2 as10 mutant embryos did not predict any changes in the amino acid sequence (data not shown). Next, we examined the noncoding regions of the mil/edg5 locus via ligation-mediated polymerase chain reaction (PCR) and identified a large insertion (more than 1,000 base pairs from linkage group 11) in the first intron of mil/edg5, near exon 1 (data not shown). To confirm this insertion, we performed PCR with three primers, including F1 and R1 outside the insertion and I1 within the insertion ( Figure S1B and S1C). A 426-bp DNA fragment amplified by primers I1 and R1 was detected only in the s1pr2 as10 heterozygote and homozygote mutant embryos. To examine the impact of the insertion on mil/edg5 gene expression, we performed quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) to compare the expression levels of mil/edg5 between the s1pr2 as10 mutants and their WT siblings during and after cardiogenesis. Significantly reduced mil/edg5 mRNA levels were detected in the s1pr2 as10 mutants prior to 96 hpf ( Figure S1D). These results strongly suggest that the s1pr2 as10 cardiac defect is caused by hypomorphic mil/edg5 gene expression and that s1pr2 as10 may represent a new allele of mil/s1pr2 m93 . Complementation experiments further demonstrated that s1pr2 as10 failed to complement a mil mutant at 24 hpf ( Figure 1L and 1M vs. 1K). In summary, s1pr2 as10 is a hypomorphic allele of mil that is caused by an insertion into intron 1.

s1pr2 as10 Mutants Exhibit a Weaker Phenotype than mil Mutants
Next, we carefully compared the cardiac phenotypes of s1pr2 as10 and mil mutations. We observed that the average distance between bilateral hearts at 24 hpf in s1pr2 as10 mutant embryos was shorter than that of mil mutants ( Figure 1J and 1K). In addition, while the bilateral hearts persisted until at least 48 hpf in mil mutants, the myocardia fused by 48 hpf in s1pr2 as10 mutants ( Figure 1F). Both of these observations indicate that the cardia bifida of s1pr2 as10 mutants is weaker than that of mil mutants. We subsequently conducted time-lapse analyses of cmlc2-stained embryos to investigate the timing of bilateral heart fusion in the s1pr2 as10 mutant ( Figure 1N). At 24 hpf, the majority of s1pr2 as10 mutant embryos contained two clusters of laterally positioned myocardial precursors, whereas by 28 hpf, over 80% of mutant embryos contained myocardial precursors in contact with one another ( Figure 1O). To demonstrate that the s1pr2 as10 mutant phenotype could be attributed to the low levels of mil/edg5 mRNA, we injected different doses of edg5-MO or edg5-5mm MO into Tg(cmlc2:EGFP, cmlc2:H2AFZmCherry) cy13 transgenic fish at the one-cell zygote stage. To quantitatively assess the severity of cardiac abnormalities, we grouped the cardiac phenotypes into three classes; Class I represented a normal heart tube, Class II represented cardiomyocytes either in close proximity or in contact, and Class III represented the most severe cardia bifida ( Figure  S2A). At 24 hpf, dose-dependent cardia bifida was detected in embryos injected with 1 to 16 ng of edg5-MO, but not in embryos injected with edg5-5mm ( Figure S2B). Injection of s1pr2 as10 mutant embryos with 30 pg mil/edg5 mRNA resulted in a partial rescue of cardia bifida at 24 hpf, as compared to mutant embryos injected with LacZ ( Figure S3). Overall, these data indicate that the hypomorphic migration defect of myocardial precursors in s1pr2 as10 mutant embryos was a result of decreased levels of mil/ edg5 mRNA.
Raising s1pr2 as10 Embryos at low Temperature Rescued the Cardiac Phenotype The hearts of s1pr2 as10 mutant embryos raised at a standard temperature (28.5uC) contained two atria and one tightly-packed ventricle, as visualized by cmlc2 and amhc double-staining; most of these embryos lacked blood circulation at 72 hpf ( Figure 2A). However, we found that the s1pr2 as10 mutants raised at a lower temperature (22.5uC) presented with a less severe phenotype ( Figure 2B). More than 85% of the s1pr2 as10 mutant embryos raised at 22.5uC had hearts with one atrium and one ventricle at 72 hpf, as compared to only 12% of the embryos raised at 28.5uC ( Figure 2C). In addition, normal blood circulation was detected at 72 hpf in 69% of the s1pr2 as10 mutant embryos raised at 22.5uC, but in only 9% of those raised at 28.5uC ( Figure 2D). Rescued s1pr2 as10 homozygous mutants grew to adulthood. A high proportion of the offspring of rescued mutants were also rescued when raised at 22.5uC (Table S1). However, the tail blister phenotype of the s1pr2 as10 mutant embryos could not be rescued by development at 22.5uC ( Figure 2B). This result indicates that incubation at low temperature specifically affects myocardial migration, and that s1pr2 as10 is not simply a temperature-sensitive allele of mil/edg5. We subsequently performed temperature shift experiments, by incubating s1pr2 as10 mutants from different stages at 28.5uC or 22.5uC. We found that the critical time window in which low temperature could rescue cardia bifida in s1pr2 as10 mutants was from ,16 to the 22 somite stage (ss) ( Figure S4).

Low Temperature Shifts the Timing of Cardiac Cone Formation
To examine why the cardiac defects of the s1pr2 as10 mutants were rescued by low temperature, we conducted time-lapse experiments to observe bilateral myocardial precursor migration at different somite stages. In Tg(cmlc2:EGFP, cmlc2:H2AFZmCher-ry) cy13 transgenic embryos raised at 28.5uC, the bilateral myocardial precursors fused at the midline and became cone-shaped at the 21-ss ( Figure 3B, red boxes in WT at 28.5uC). Surprisingly, the timing of cardiac cone formation was shifted to the 19 ss when the embryos were raised at the lower temperature ( Figure 3B, red boxes in WT at 22.5uC). Although the overall growth rate at 22.5uC was about 2-fold slower than that at 28.5uC [27], low temperature does not seem to cause any morphological defects in WT and s1pr2 as10 embryos ( Figure 3A). Similarly, the bilateral cardiomyocytes in the s1pr2 as10 mutant embryos fused at the 26 ss at low temperature ( Figure 3B, red boxes in s1pr2 as10 22.5uC); this result was in contrast to the delayed fusion (28 hpf) observed at 28.5uC ( Figure 1O). In situ hybridization using a cmlc2 probe revealed that 77% WT and 84% s1pr2 as10 mutant embryos raised at low temperature formed a fused cardiac cone at the 19 ss and 26 ss, respectively ( Figure 3C). In contrast, the majority of WT and s1pr2 as10 mutant embryos raised at standard temperature presented with horseshoe-shaped hearts or bilateral hearts at 19 ss and 26 ss, respectively ( Figure 3C). However, cardia bifida of mil mutant embryos cannot be rescued by low temperature treatment ( Figure S5A). Since defects in endoderm formation also result in cardia bifida [7,28], we induced cardia bifida by injecting different doses of gata5 MO or bon MO. We found that the cardia bifida phenotypes of embryos injected with 10 ng of gata5 MO or bon MO can be significantly rescued by being raised at low temperature ( Figure 3D and 3E) while cardia bifida defects in embryos injected with 20 ng of gata5 MO were not rescued by low temperature treatment ( Figure S5B). Although the rescue effect was not statistically significant, a reduced percentage of cardia bifida phenotype was notable in embryos injected with 20 ng of bon MO. (Figure S5B). In addition to cardia bifida, cells of the anterior gut tissues also failed to migrate to the midline in gata5/fau mutants [28]. We found that the distance between the two lateral anterior gut tubes was also significantly reduced in embryos injected with 10 ng gata5 MO and incubated at low temperature, as compared to those incubated at standard temperature (145 mm vs. 160 mm) at the prim-25 stage ( Figure S6). Thus, low temperature alters the timing of cardiac cone formation in WT, s1pr2 as10 mutants (which exhibit hypomorphic migration defects) and other cardia bifida morphants, resulting in a rescue of the cardia bifida phenotype that is independent of the genetic background.

Gene Expression Profiles of Zebrafish Embryos Raised at 28.5 and 22.5uC
Previous reports have demonstrated that changes in temperature affect developmental processes and gene expression in zebrafish embryos [29,30]. We performed DNA microarray ( Figure 4) and digital gene expression (DGE) ( Table S2) analyses to further investigate how low temperature mitigates cardia bifida. By combining these two analyses, we identified several candidate gene groups and demonstrated that genes encoding extracellular matrix (ECM) proteins are important for preventing cardia bifida. DNA microarray and qRT-PCR revealed that mil/edg5 levels were not affected by raising WT or s1pr2 as10 embryos at different temperatures (28.5uC or 22.5uC) ( Figure S7).

Low Temperature Mitigates Cardia Bifida by Altering ECM Gene Expression
Based on the DNA microarray and DGE analyses, we established that several genes involved in the ECM were affected by low temperature; these genes included tenascin-c (tnc), tenascin-w (tnw), and collagen type II, alpha-1a identified through the DNA microarray analysis (Figure 4), and fibronectin 1 (fn1), collagen, type XI alpha 2 and sid4 through the DGE analysis (Table S2). We first focused on fn1, which was previously found to be important for myocardial migration [8]. An increased level of fn1 was initially identified via DGE in s1pr2 as10 mutant embryos raised at 22.5uC (Table S2). Whole mount in situ hybridization and immunocytochemistry revealed increased expression of fn1 mRNA and protein in the midline regions and bilateral anterior LPM of both 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC ( Figure 5B and 5D and Figure S8). However, levels of fn1 at the tail bud or yolk regions were not affected by incubation at low temperature ( Figure 5E-5H). qRT-PCR revealed an approximately 1.3-fold increase of fn1 mRNA in s1pr2 as10 mutant embryos raised at 22.5uC, as compared to those raised at 28.5uC. However, this temperature-mediated up-regulation was not statistically significant ( Figure 5I). The failure to detect a significant change in WT and s1pr2 as10 mutant embryos raised at low temperature may be due to the inability of temperature to affect fn1 expression at the tail bud and yolk regions ( Figure 5E-5H).
As incubation at low temperature raised fn1 expression and rescued the cardia bifida phenotype in s1pr2 as10 , we hypothesized that fnl up-regulation may be required for the rescue of heart development. To test this hypothesis, we injected fn1-MO into s1pr2 as10 mutants raised at 22.5uC. As shown in Figure 5J, 26-ss mutant embryos injected with 1.25 or 2.5 ng fn1-MO displayed severe cardia bifida, even after being raised at the low temperature. In addition, we injected human fibronectin protein into the midline region ( Figure 5K') of 14-16 hpf s1pr2 as10 mutant embryos raised at 28.5uC, to determine whether elevated levels of fibronectin can rescue the cardia bifida phenotype. We found that the percentage of mutant embryos with class III severe cardia bifida was reduced from 60% to 42% ( Figure 5K). Together, these results indicate that incubation at low temperature mitigates cardia bifida in s1pr2 as10 mutants by promoting fn1 expression in the midline and bilateral anterior LPM regions.
Two Tenascin Genes, Tenascin-C and Tenascin-W, Play Important Roles in myocardial migration in embryos raised at low temperature In addition to fn1, expression levels of two other extracellular matrix genes, tenascin-C (tnc) and tenascin-W (tnw), were affected by incubation at low temperature. The tnc gene can produce four different isoforms through differential splicing. By using RNA probes against exons common to all tnc isoforms for in situ hybridization, we detected increased expression of tnc at the pharyngeal arches and in the cells between the brain and eyes of 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC as compared to those raised at 28.5uC ( Figure 6B and 6D vs. 6A and 6C). However, expression of tnc in the somite region was not affected by the temperature of incubation ( Figure 6E-6H). Accordingly, the use of qRT-PCR to measure tnc mRNA did not uncover significant differences between WT and s1pr2 as10 mutant embryos raised at 22.5uC as compared to those raised at 28.5uC ( Figure 6I). We subsequently used a previously published tnc-MO1 that can prevent the translation of all tnc isoforms [22] and designed a tnc25mm MO based on the tnc-MO1 sequence for use as controls. Additionally, we used the previously published tnc-MO2, which blocks splicing of tnc exon 1 [22]. Mutant s1pr2 as10 embryos incubated at 22.5uC and injected with either tnc-MO1 or tnc-MO2 presented with more severe cardia bifida phenotypes, as compared to tnc25 mm injected embryos ( Figure 6J and Figure  S9A). We also injected human TNC protein into the midline region of s1pr2 as10 mutants, and found the percentage of embryos with class III severe cardia bifida defects was reduced from 64% to 33% ( Figure 6K). These results indicate that low temperature rescues s1pr2 as10 cardiac phenotype partially due to increased expression of tnc.
DNA microarray analysis also revealed decreased expression of another tenascin gene, tnw, in 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC. Whole-mount in situ hybridization and qRT-PCR confirmed that tnw mRNA expression was reduced in all tissues, including the LPM, of both 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC ( Figure 6M, 6O and 6P). We subsequently designed morpholinos (tnw-MO1 and tnw-MO2) against the 59 untranslated region (UTR) of tnw to prevent its translation; we also designed a tnw25 mm based on the tnw-MO1 sequence as a control. In order to test the efficiency and specificity of tnw-MOs, we fused EGFP in frame with the tnw 59UTR under the control of the CMV enhancer/promoter, and co-injected this construct with tnw-MO1, tnw-MO2, or tnw25 mm into one-cell zygotes. As shown in Figure S9D and E, we detected EGFP expression in the majority of embryos co-injected with tnw25 mm and CMV-tnwUTR-EGFP plasmid. In contrast, the majority of embryos (85%) co-injected with tnw-MO1 and CMV-tnwUTR-EGFP did not express GFP. However, 30% of embryos co-injected with tnw-MO2 and CMV-tnwUTR-EGFP expressed GFP, indicating that tnw-MO1 is more efficient at blocking EGFP translation.
This may be a consequence of tnw-MO2 targeting a region further upstream from that of tnw-MO1.
We then injected either tnw-MO1 or tnw-MO2 into s1pr2 as10 mutants raised at 28.5uC, and observed partial rescue of the cardia bifida phenotype at 24 hpf ( Figure 6Q and Figure S9B). Injection of tnw-MO1 resulted in greater rescue of the cardia bifida phenotype than with tnw-MO2; this may be attributed to the reduced efficiency of tnw-MO2 in blocking translation. Furthermore, some of the rescued embryos contained a single heart tube and displayed normal circulation (data not shown). Interestingly, injection of tnw mRNA at the one-cell stage caused a highly severe cardia bifida phenotype in s1pr2 as10 mutants, which was observed even at 22.5uC ( Figure 6R).
We also injected fn1-MO, tnc-MO1, or tnw mRNA into wild type embryos to investigate their effects on myocardial migration. We found that the percentage of tnc-MO1-injected morphants (26%) which developed a horseshoe-shaped heart at 22ss was similar to that of control embryos (27%) (Figure S10A and S10B). The percentage of tnw mRNA-injected embryos with a horseshoe-shaped heart at 22ss was greater (38%) than that of control, and 12% of these embryos also exhibited cardia bifida ( Figure S10D). The percentage of embryos with horseshoeshaped hearts at 22 ss was increased further by fn1-MO1injection (69%), and 9% of these embryos exhibited cardia bifida (9%) ( Figure S10C); these results indicate that fibronectin plays a more important role in regulating myocardial migration than tenascin-C or tenascin-W. However, the majority (93%) of Together, these data indicate that tnc and tnw play opposing roles in zebrafish myocardial migration, and that low temperature mitigates cardia bifida by altering the expression of ECM genes: fn1 and tnc are up-regulated, and tnw is down-regulated.

Reactive Oxygen Species (ROS) Mediate Mitigation of Cardia Bifida by Low Temperature
Cold exposure was previously shown to increase expression of peroxisome proliferator-activated receptors (PPARs) and uncoupling proteins (UCPs) in the zebrafish brain, to prevent oxidative damage and to maintain metabolic balance and homeostasis [31].
We investigated whether ROS mediates mitigation of the cardia bifida phenotype in s1pr2 as10 mutants by low temperature treatment. To test this hypothesis, we treated s1pr2 as10 mutant embryos with either 50 or 150 mM of N-acetyl cysteine (NAC), a ROS scavenger, at the tailbud stage, and incubated the embryos at 22.5uC. As shown in Figure 7, we detected a dose-dependent of NAC on the percentage of s1pr2 as10 mutant embryos with a class I single heart tube at 26 ss when incubated at low temperature; class I heart tubes were reduced by NAC to 46-32% as compared to 67-73% in untreated s1pr2 as10 mutant embryos. This result suggests that ROS may mediate mitigation of cardia bifida by low temperature.

Discussion
Previous reports have demonstrated that changes in temperature affect various aspects of development and gene expression [29,30]. In this study, we have demonstrated that temperature also affects the timing of cardiac cone formation. We have specifically demonstrated that low temperature (22.5uC) mitigates cardia bifida in both s1pr2 as10 mutant and gata5 and bon morphant embryos. DNA microarray and digital gene expression analyses revealed that low temperature regulates expression of several extracellular matrix (ECM) genes to control myocardial migration.
An increase in fn1 expression at the midline region and bilateral anterior lateral plate mesoderm (LPM) was identified in both 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC ( Figure 5B, and 5D and Figure S8). Injection of s1pr2 as10 mutant embryos with fn1-MO abolished the rescue of cardia bifida by low temperature (Figure 5J), while injection with human fibronectin protein decreased the percentage of s1pr2 as10 mutant embryos raised at 28.5uC with class III severe cardia bifica at 24 hpf ( Figure 5K). During myocardial precursor migration, fibronectin secreted by the endoderm and/or endocardial precursors is deposited at the midline and laterally around the myocardial precursors, to maintain the integrity of myocardial epithelia, and mediate interactions between the endoderm and migrating myocardial precursors [8,15]. It has been demonstrated that knockdown of mil/edg5 results in cardia bifida and decreased levels of fibronectin [11]. Furthermore, injection of the midline with fibronectin partially rescued the cardia bifida phenotype in mil mutant embryos [15]. All of these findings suggest that both mil/edg5 and low temperature play a positive role in regulating myocardial migration. Proper deposition of extracellular matrix proteins is also important for normal cardiac function after formation of the four-chambered heart, and excess collagen and fibronectin deposition results in cardiac fibrosis in cardiac hypertrophy. Reactive oxygen species (ROS) have been shown to up-regulate expression of profibrotic genes, such as col2a1, col1a1, and fn1, thereby resulting in direct extracellular matrix deposition in a transgenic mouse model, and resulting in cardiac hypertrophy [32].
Tenascins are a family of large multimeric extracellular matrix glycoproteins involved in cell adhesion, migration and proliferation [33,34]. Tenascin-C has been shown to either promote or inhibit cell adhesion and migration, depending on the cell type. Addition of tenascin-C stimulated the migration of human adult dermal fibroblasts cultured within matrices comprised of fibro-nectin [35]. Smooth muscle cells cultured on tenascin-C substrate exhibited increased expression of av-integrin and elevated phosphorylation of focal adhesion kinase (FAK), followed by enhanced migration [36]. Tenascin-C also promoted bovine retinal endothelial cell migration by enhancement of FAK phosphorylation [37]. On the other hand, high levels of tenascin-C produced by glioma tumour tissues actively hindered T-cell migration, resulting in an accumulation of T cells in the peritumoral region [38]. Although the function of tenascin-W and its gene regulation has not been well-characterized, it has been shown to confer anti-adhesive properties [39]. Addition of soluble tenascin-W was reported to inhibit the formation of lamellipodia with stress fibers and focal adhesion complexes in a murine myoblast cell line (C2C12) cultured on fibronectin. Various factors, including inflammatory cytokines, growth factors, oxidative stress, and hypoxia, have been shown to induce tnc expression [40]. In neonatal rat cardiomyocytes, mechanical strain increased tnc expression by activating nuclear factor-kB through ROS [41]. Here, we observed up-regulation of tnc and down-regulation of tnw in embryos raised at low temperature ( Figures 5 and 6). MO knockdown and mRNA or protein rescue confirmed the opposing roles of tenascin-C and tenascin-W in mitigating cardia bifida in zebrafish embryos. Moreover, we found that the addition of Nacetyl cysteine (NAC), a ROS scavenger, decreased mitigation of cardia bifida in s1pr2 as10 mutant embryos raised at 22.5uC ( Figure 7A).
Based on our results and previously published studies, we propose a model describing how low temperature mitigates cardia bifida in zebrafish embryos ( Figure 7B). Low temperature may enhance ROS production in cranial neural crest cells, endocardial and myocardial cells, and endoderm (in a similar manner to ROS induction in zebrafish brain upon cold exposure [31]), as these tissues are more vulnerable to oxidative stress [42][43][44][45][46][47]. The presence of ROS in the endoderm and anterior LPM may then up-regulate fn1 expression, resulting in the secretion of fibronectin in the midline and bilateral anterior LPM ( Figure 5 and Figure  S8). Meanwhile, ROS production in the cranial neural crest derived-bilateral pharyngeal arches may activate tnc expression, thereby promoting secretion of tenascin-C near migrating bilateral cardiomyocytes ( Figure 6). Increased levels of secreted tenascin-C may enhance integrin expression and promote FAK phosphorylation in migrating bilateral cardiomyocytes, thereby facilitating their interaction with fibronectin, formation of focal adhesion complexes, actin polymerization, and protrusion [37,[48][49][50]. Increased fibronectin around migrating cardiomyocytes can further promote the integrity of myocardial epithelia, which is a prerequisite of myocardial migration [8]. Since tenascin-W is known to inhibit cell adhesion to fibronectin, ROS may also down-regulate tnw expression, consequently preventing secretion of tenascin-W from scattered epidermal cells in the head region, and thereby allowing migrating cardiomyocytes to interact with Figure 6. Expression of two tenascin genes is affected by temperature. In situ hybridization against tenascin-c (tnc) (A-H) and tenascin-w (tnw) (L-O) in 22-ss wild-type (WT) and s1pr2 as10 (as10) mutant embryos raised at 28.5uC or 22.5uC. Expression of tnc increased in the pharyngeal arches (white arrows in B and D) and the cells between the brain and eyes of embryos raised at 22.5uC, but was unaffected in the trunk somite region (E-H). (I) qRT-PCR revealed a trend towards increased expression of tnc mRNA in s1pr2 as10 mutant embryos raised at 22.5uC. (J) Knockdown of tnc expression with tnc-MO1 in s1pr2 as10 mutants increased the percentage of 26-ss embryos raised at 22.5uC with the Class III cardia bifida phenotype. (K) Injection of s1pr2 as10 mutant embryos raised at 28.5uC with human TNC protein partially rescued cardia bifida phenotypes at 24 hpf. (L-O) Decreased tnw expression in all tissues, including scattered epidermal cells in the head (black arrows in M and O) was observed in both 22-ss WT and s1pr2 as10 mutants raised at 22.5uC. (P) qRT-PCR was used to confirm that expression of tnw mRNA is significantly reduced in 22-ss WT and s1pr2 as10 mutant embryos raised at 22.5uC. (Q) Knockdown of tnw with tnw-MO1 in s1pr2 as10 mutant embryos raised at 28.5uC partially rescued cardia bifida phenotypes at 24 hpf. (R) Injection of s1pr2 as10 mutant embryos raised at 22.5uC with 100 pg tnw mRNA increased the percentage of 26-ss embryos with the Class III cardia bifida phenotype. Scale bars = 100 mm. The error bars indicate the standard error. Statistical significance was determined using Student's t-test. * indicates p,0.05, ** indicates p,0.01, *** indicates p,0.001. doi:10.1371/journal.pone.0069788.g006 fibronectin [51]. Furthermore, a combinatorial effect of upregulation of tnc and fn1, and down-regulation of tnw are required to mitigate cardia bifida of s1pr2 as10 mutant embryos and gata5 and bon cardia bifida morphants upon low temperature treatment.
In conclusion, our identification of the s1pr2 as10 mutation serendipitously led to our finding that incubation at low temperature mitigates cardia bifida, a phenomena caused by upregulation of fn1 and tnc and down-regulation of tnw. In addition, ROS mediates mitigation of cardia bifida upon low temperature treatment. Future studies may be expected to delineate the mechanisms underlying regulation of fn1, tnc and tnw expression by ROS under low temperatures. Figure S1 Genetic mapping of the s1pr2 as10 mutant gene. (A) Meiotic map of the s1pr2 as10 locus. The numbers above the line indicate the number of meiotic recombination events out of the total number of meioses. (B) An enlarged view of the mil/ edg5 locus shows the insertion in intron 1. Three primers (F1, R1, and I1) were used to confirm the insertion. (C) A 182-bp DNA fragment amplified using primers F1 and R1 was detected in wildtype (WT) and s1pr2 as10 heterozygous mutants, whereas a 426-bp DNA fragment amplified using primers I1 and R1 was detected in s1pr2 as10 heterozygous and homozygous mutant embryos. (D) Expression of mil/edg5 mRNA was reduced in s1pr2 as10 mutants prior to 96 hpf, as revealed by qRT-PCR. WT (+, W); s1pr2 as10 mutant (2, M). The error bars indicate the standard error.  Table S1 Circulation defect of s1pr2 as10 mutant embryos from different genotypes could be partially rescued when raised at 22.5uC. The statistics of the offspring derived from different genotypes of s1pr2 as10 mutant raised at 28.5uC or 22.5uC. Genotype labeled with +/2 or 2/2 indicated the heterozygous or homozygous s1pr2 as10 mutants respectively. For example, among 213 embryos from heterozygous mutants intercross (+/2 x +/2) raised at 28.5uC, 162 embryos showed normal wild-type phenotype (a), 46 embryos contained tail blisters phenotype and established no blood circulation (b), and 5 embryos contained tail blisters with normal circulation (c). Mendel ratio was calculated by number of embryos with tail blister phenotype divided by number of total embryos. Rescue of tail blister with circulation phenotype was observed in offspring derived from different genotypes of s1pr2 as10 mutant raised at 22.5uC. Rescue percentage of tail blister with circulation phenotype was calculated by number of embryos showing tail blister with circulation phenotype divided by total number of embryos showing tail blister phenotype. (DOC) Table S2 Up-and down-regulated genes in s1pr2 as10 mutant at 22.5uC as determined by digital gene expression (DGE) analysis. Double-stranded cDNA from 22 ss-s1pr2 as10 mutant embryos raised at 28.5uC and 22.5uC were synthesized for next generation sequencing. The Up-and downregulated genes at 22.5uC were subsequently analyzed by pathway analysis according to Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Several groups of genes were selected for further analysis. (DOC)